Skip to content

clanker: revenue is 5x too high for the 263 days before the 20% fix (#3760) #8317

Description

@mkzung

clanker revenue is 5x too high for the 263 days before the 20% fix

fees/clanker.ts used to report every dollar of fees as revenue. #3760 corrected that to
20% on 2025-07-29 and the adapter has been right ever since, but the stored history was
never regenerated, so everything before that date still carries the old 100% number.

The switch is visible to the day, with no transition period:

2025-07-26     $23,781      $23,781   1.0000
2025-07-27     $31,027      $31,027   1.0000
2025-07-28     $74,775      $74,775   1.0000
2025-07-29     $79,049      $15,810   0.2000   <- #3760 merged 11:29 UTC
2025-07-30    $129,773      $25,955   0.2000
2025-07-31    $342,726      $68,545   0.2000

which lines up with what #3760 changed:

-  dailyRevenue: dailyFees,                 Revenue: "All fees are collected by Clanker protocol."
+  const dailyProtocolRevenue = dailyFees.clone(0.2)
+  dailyRevenue: dailyProtocolRevenue,      Revenue: "Clanker protocol collects 20% of LP fees."

Over the stale window:

2024-11-08 .. 2025-07-28        263 days
fees                            $21,992,943
published revenue               $21,992,943   ratio exactly 1.000000
at the 20% the adapter now uses   $4,398,589
overstated by                   $17,594,354

$21,992,943 * 0.2 = $4,398,589, and $21,992,943 - $4,398,589 = $17,594,354.

All-time revenue would go from $31,198,622 to $13,604,268.

Supply-side is the other half of the same gap. It is $0 across the whole window, where 80%
of fees is $17,594,354.

Separately, the stored history contradicts the adapter's own code. clanker.ts today
returns dailyProtocolRevenue: dailyRevenue, the same object, so the two can never differ.
In the window they differ on 201 of 263 days, and protocolRevenue is simply absent until
2025-05-28. Total gap $20,691,274. That part needs no view on what the right fee split is,
it is just the stored series disagreeing with the code that is supposed to produce it.

A refill works here. The adapter is not runAtCurrTime, it has start: "2024-11-08", and
it reads WETH received by the fee wallets through addTokensReceived, so re-running the
window regenerates it from chain data rather than needing a source that no longer exists.

https://api.llama.fi/summary/fees/clanker?dataType=dailyFees
https://api.llama.fi/summary/fees/clanker?dataType=dailyRevenue
https://api.llama.fi/summary/fees/clanker?dataType=dailyProtocolRevenue
https://api.llama.fi/summary/fees/clanker?dataType=dailySupplySideRevenue

One caveat I would rather state than leave implied. I am treating the 20% as correct
because that is what the adapter and its methodology string both say now. If clanker really
did take 100% of fees before 2025-07-29 and only moved to 20% later, then the history is
right and this is not a bug. What makes that unlikely is the timing: the change lands on
the exact day an adapter edit merged, not on a protocol announcement.

Same shape as #8294, #8302 and #8312.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions